STATS 744: Homework 6

About the data

This a five-year survival rates data for the patients of various Cancer types in the US from \(1963\) to \(2013\). The data was published by the US National Cancer Institute and downloaded from Data World.

Aim

• Show trends in cancer survival over the five-year periods
  • By gender
  • By race

Download and load data

## Create output directory
createDir(dirname = "HW6", dirpath = ".")

## Load/download data
url = "https://query.data.world/s/idhkwmkwynt4n4jlgmkslf43c5pd24"
raw_df <- downloadDf(filename = "cancer_survival_rates_usa"
    , filetype = "csv"
    , df_url = url
)

Reading dataset from your computer… cancer_survival_rates_usa.csv dataset already saved!!!

Variables

• The table below summarizes the variables in the dataset

df_summary <- (raw_df
    %>% mutate_at("Year", as.factor)
    %>% summarizeDf()
    %>% setnames("Summary", "Summary ([min, max]; mean (sd) / label(%))")
)
datatable(df_summary, rownames = FALSE)

Cleaning

• Drop all categories containing all the other subcategories:
  • All races
  • total
  • All cancers
• Rename the variables
• Drop cases with missing survival rates

# To plot overall cancer survival for all cancer types
raw_df <- (raw_df
    %>% filter(!is.na(Survival.Rate))
    %>% setnames(names(.), tolower(gsub("\\.", "_", names(.))))
)

# To plot cancer survival rates by race and gender
working_df <- (raw_df
    %>% filter(!grepl("^All", race) & !grepl("All", cancer_type) & !grepl("total", gender))
)

Visualization

Overall Cancer Survival

• The plot below shows the overall five-year cancer survival rates in the US. The rates disregards gender and race segregation.
• Generally, prostrate, thyroid, skin, and breast cancer have the highest survival rates while pancreas, liver, lung and esophagus have the lowest.

overall_df <- (raw_df
    %>% filter(grepl("^All", race) & grepl("total", gender) & !grepl("^All", cancer_type))
    %>% group_by(year)
    %>% mutate(ranking = min_rank(-survival_rate))
)

plot1 <- (ggplot(overall_df, aes(ranking, group = cancer_type, colour = cancer_type, fill = cancer_type))
    + geom_tile(aes(y = survival_rate/2, height  = survival_rate, width = 0.9), colour = NA)
    + geom_text(aes(y = 0, label = paste0(cancer_type, " ")), vjust = 0.2, hjust = 1)
    + geom_text(aes(y = survival_rate, label = paste0(" ", percent(survival_rate)), hjust = 0))
    + coord_flip(clip = "off", expand = TRUE)
    + scale_x_reverse()
    + scale_color_viridis_d(name="")
    + scale_fill_viridis_d(name="")
    + guides(color = FALSE, fill = FALSE)
    + theme_minimal()
    + theme(plot.title = element_text(hjust = 0.5, face = "bold", colour = "grey")
        , axis.ticks.y = element_blank()
        , axis.text.y = element_blank()
        , plot.margin = margin(1,1,1,4, "cm")
    )
    + transition_states(states = year, transition_length = 4, state_length = 1)
    + ease_aes("cubic-in-out")
    + labs(title = "Cancer survival rate per Year : {closest_state}"
        , x = ""
        , y = "Survival rate"
    )
)
animate(plot1
    , nframes = 100
    , fps = 20
    , width = 950
    , height = 750
    , end_pause = 10
)

anim_save("csurvival_overall.gif", plot1)

Cancer Survival by Gender and Race

• The figure below shows the cancer survival rates compared across race and gender.

plot2 <- (ggplot(working_df, aes(x = year, y = survival_rate, colour = cancer_type, group = cancer_type))
    + geom_line()
    + scale_x_continuous(breaks = seq(1963, 2013, 10))
    + scale_y_continuous(labels = percent)
    + geom_point()
    + geom_text(aes(y = survival_rate, label = cancer_type))
    + scale_color_viridis_d(name="")
    + facet_grid(gender ~ race)
    + transition_reveal(year)
    + theme(legend.position="none")
)
animate(plot2
    , nframes = 100
    , fps = 20
    , width = 950
    , height = 750
    , end_pause = 10
)

anim_save("csurvival_racegender.gif", plot2)